ME41006 · Learning notes · 02

Image Lab

Inspect the relationship between camera pixels, a binary mask, and the location inferred from that mask.

01

From pixels to a geometric estimate

Each lesson asks a different question about what the camera sees.

  1. 01

    Find the Color

    Select yellow pixels and inspect the binary mask.

  2. 02

    One Plate

    Return an aim point that belongs to one plate.

  3. 03

    Plate Center

    Estimate a plate’s center when part of its outline is occluded.

02

Read the perception output

The three monitors separate the input image from the result returned by your code.

Image Lab with Capture, Mask and Result monitors beneath the scene
Find the Color · starter output · seed 7.Open image ↗
1

Capture

The camera image is the input. It stays fixed between runs on this pile.

2

Mask

White pixels are selected by the program. Black pixels are excluded.

3

Result

Selected pixels retain their original color, making missing or unwanted regions easier to inspect.

03

Separate selection from localization

Occlusion changes the visible shape of a plate.

Plate Center lesson showing an occluded plate, mask and aim point evaluation
Plate Center · starter output · seed 7.Open image ↗
1

Visible shape

The camera records only the exposed portion of the plate.

2

Selected region

The mask represents the visible region returned by the program.

3

Aim point

The rings show how close the aim point is to the plate’s center.

04

Questions for your experiment

Compare the same capture first, then test with a new pile or layout.

SELECTIONWhich target pixels are missing, and which unwanted pixels are included?

LOCALIZATIONDoes the aim point refer to one object and the location the task asks for?

GENERALIZATIONDoes the result remain useful when the pile, layout, or visible shape changes?